Optimized Long-Short Term Memory Approach using Partial Opposition-based Particle Swarm Optimization for Diabetes Detection

It has always been challenging to create a diagnosis system to effectively identify diabetes in the context of e-healthcare. This study proposes a new framework to analyse medical data for disease detection, wherein the Long-Short-Term Memory (LSTM) techniques are playing an increasingly important role in healthcare services. Although the LSTM has numerous benefits, choosing the hyperparameters can be challenging, much like with conventional neural networks. This study has effectively addressed this problem by incorporating a LSTM model optimization based on Partial Opposition based Particle Swarm Optimization (POPSO). POPSO is used to optimize the parameters of LSTM since it is difficult to choose the LSTM’s hyperparameters. The experimental findings demonstrate that the optimized model performs better and yields a more precise diabetes prediction